Molecular Cell

Article
Substrate-Assisted Catalysis by PARP10 Limits Its Activity to Mono-ADP-Ribosylation
Henning Kleine,1 Elzbieta Poreba,1,4 Krzysztof Lesniewicz,1,5 Paul O. Hassa,2,6 Michael O. Hottiger,2 David W. Litcheld,3 Brian H. Shilton,3,7 and Bernhard Lu scher1,7,*
r Biochemie und Molekularbiologie, Klinikum, RWTH Aachen University, 52057 Aachen, Germany fu of Veterinary Biochemistry and Molecular Biology, University of Zurich, 8057 Zurich, Switzerland 3Department of Biochemistry, University of Western Ontario, London, ON N6A 5C1, Canada 4Present address: Institute of Experimental Biology, A.M. University, Miedzychodzka 5, 60-371 Poznan, Poland 5Present address: Institute of Molecular Biology and Biotechnology, A.M. University, Miedzychodzka 5, 60-371 Poznan, Poland 6Present address: European Molecular Biology Laboratory, Gene Expression Unit, Meyerhofstrasse 1, 69117 Heidelberg, Germany 7These authors contributed equally to this work *Correspondence: luescher@rwth-aachen.de DOI 10.1016/j.molcel.2008.08.009
2Institute 1Institut

SUMMARY

ADP-ribosylation controls many processes, including transcription, DNA repair, and bacterial toxicity. ADPribosyltransferases and poly-ADP-ribose polymerases (PARPs) catalyze mono- and poly-ADP-ribosylation, respectively, and depend on a highly conserved glutamate residue in the active center for catalysis. However, there is an apparent absence of this glutamate for the recently described PARP6PARP16, raising questions about how these enzymes function. We nd that PARP10, in contrast to PARP1, lacks the catalytic glutamate and has transferase rather than polymerase activity. Despite this fundamental difference, PARP10 also modies acidic residues. Consequently, we propose an alternative catalytic mechanism for PARP10 compared to PARP1 in which the acidic target residue of the substrate functionally substitutes for the catalytic glutamate by using substrate-assisted catalysis to transfer ADP-ribose. This mechanism explains why the novel PARPs are unable to function as polymerases. This discovery will help to illuminate the different biological functions of mono- versus polyADP-ribosylation in cells.
INTRODUCTION The modication of proteins by ADP-ribosylation is a phylogenetically ancient mechanism. Mono-ADP-ribosylation was originally identied as a pathogenic mechanism of certain bacterial toxins (Aktories and Barbieri, 2005; Holbourn et al., 2006). Mammalian cells contain ectoenzymes with mono-ADP-ribosyltransferase (mART) activity, while the molecular identity of postulated intracellular enzymes is less clear (Corda and Di Girolamo, 2003; Hassa et al., 2006). Recent evidence suggests that mitochondrial SIRT4 and nuclear SIRT6 possess mART activity (Haigis et al., 2006). While mARTs transfer single ADP-ribose units, a different class of enzymes, poly-ADP-ribose polymerases (PARPs),

is responsible for poly-ADP-ribosylation of protein substrates (DAmours et al., 1999; Schreiber et al., 2006). PARPs regulate many cellular activities, including DNA repair, apoptosis, and chromatin dynamics (DAmours et al., 1999; Kim et al., 2005; Schreiber et al., 2006), and PARP inhibitors have proven successful in cancer therapy (Haince et al., 2005). PARP1, the founding and best-studied member of the PARP family, was for a long time considered to be the only enzyme in mammalian cells that could generate poly-ADP-ribose (pADPr) polymers. However, in recent years, additional enzymes have been described and systematic studies have indicated that 17 different proteins exist that share a PARP domain and thus may contain polymerase activity (Ame et al., 2004; Otto et al., 2005). Besides the PARP domain, a wide spectrum of protein domains are present in the different members of the PARP family, suggesting that these proteins are involved in a broad spectrum of physiological activities. The enzymatic activity of mARTs and PARPs was described to be dependent on a well-conserved catalytic glutamate residue in the active center (Bellocchi et al., 2006; Holbourn et al., 2006; Marsischky et al., 1995; Ruf et al., 1998). However, in several of the novel postulated PARP enzymes, the existence of this important glutamate is disputed by sequence comparisions, suggesting that either these enzymes are catalytically inactive or that another mechanism is in place (Ame et al., 2004; Otto et al., 2005). Thus, these ndings indicate that the PARP family may not be homogenous regarding its catalytic activity. Instead it is possible that distinct subfamilies exist that may differ in their enzymatic activity as well as in other features. PARP10 was identied as an interaction partner of the oncoprotein MYC (Yu et al., 2005) and is one of these novel PARP family members that apparently lacks the catalytic glutamate. Nevertheless, this enzyme is functional in that it auto-ADPribosylates and modies core histones (Chou et al., 2006; Yu et al., 2005). These surprising results prompted us to dene the catalytic activity of PARP10 in more detail, and we have found that it does not function as a polymerase. Instead, PARP10 is a mART that automodies glutamate 882, thereby dening a PARP subfamily that is unable to form polymers. Molecular modeling suggests that the enzymes of this family use

Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc. 57

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

Figure 1. PARP10 Functions as Mono-ADP-Ribosyltransferase

(A) C-TAP-PARP10 and C-TAP were puried from stably transfected HEK293 cells and analyzed by Coomassie blue staining.

58 Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc.

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

substrate-assisted catalysis, in which the acidic acceptor amino acid is also used to activate the NAD+ cosubstrate. This represents a mART within the PARP family. RESULTS AND DISCUSSION PARP10 Functions as Mono-ADP-Ribosyltransferase Our previous studies had shown that the C-terminal half of PARP10 possesses catalytic activity (Yu et al., 2005). The automodied PARP10 reacted weakly with mAb 10H that was reported to recognize pADPr chains of at least 20 ADP-ribose moieties (Kawamitsu et al., 1984). We concluded from this nding that PARP10 possesses polymerase activity, although we did not observe any obvious change in mobility on SDS-PAGE as has been described for PARP1 (Yu et al., 2005). This prompted us to evaluate the enzymatic properties of PARP10 in more detail. We puried full-length PARP10 and a catalytically inactive mutant (PARP10-G888W) by tandem afnity purication (TAP) from HEK293 cells (Figure 1A and data not shown). Mass spectrometry analysis and reactivity with PARP10-specic antibodies conrmed that these proteins are PARP10 (data not shown). ADP-ribosylation assays demonstrated robust C-TAP-PARP10 automodication with incorporation being linear over at least 20 min (see Figure S1A available online). However, C-TAPPARP10 did not display a mobility shift upon auto-ADP-ribosylation, which is indicative of the pADPr formation catalyzed by PARP1, PARP2, and the Tankyrases (PARP5a and PARP5b) (Figure 1 and Figure S1) (Ame et al., 1999; Naegeli et al., 1989; Rippmann et al., 2002; Smith et al., 1998). Therefore, we compared the activities of PARP10 and PARP1. Puried C-TAPPARP10 and N-TAP-PARP10 underwent automodication, while C-TAP-PARP10-G888W was inactive (Figure 1B). Importantly, neither C-TAP-PARP10 nor N-TAP-PARP10 revealed any change in mobility, unlike PARP1, which showed a substantial increase in apparent molecular weight due to pADPr formation (Figure 1B). Furthermore, and in contrast to PARP1, automodied PARP10 was not or only poorly stained by mAb 10H (Figure 1C and a 20-fold longer exposure in Figure S1B). A weak signal was measurable by pAb 96-10-04, which is described to recognize oligomers of six or more ADPr moieties (Alexis Biochemicals) (Figure 1C). In both situations, no shift in apparent mobility of PARP10 was detected. To exclude interference of the residual TAP tag with catalytic activity, we expressed HA-tagged versions of PARP10 and PARP1 transiently in HeLa cells. TAP and HA tags display no sequence

homology. Analysis of the HA-tagged proteins supported the results obtained for TAP-tagged PARP10 (Figure 1D). Finally, endogenous PARP10 from U937 promyelocytes was immunoprecipitated using a polyclonal PARP10 antiserum (Yu et al., 2005). Similarly, this protein automodied without any change in mobility (Figure 1E) and was not reactive to mAb 10H (data not shown). Thus, full-length PARP10 is unable to generate long polymers that would alter the mobility of the protein on SDS gels. The efcient catalysis of long, branched pADPr by PARP1 is stimulated by its interaction with DNA (DAmours et al., 1999). Since PARP10 possesses potential RNA and/or DNA interaction motifs (Yu et al., 2005), we tested whether nucleic acids could modify its catalytic activity. The addition of sheared salmon sperm DNA, linear or supercoiled plasmid DNA, or total cellular RNA neither stimulated the activity of full-length PARP10 nor resulted in polymer formation (Figure S1C and data not shown). Instead, DNA led to a dose-dependent reduction of automodication. This was also true for GST-PARP10(8181025), which contains the catalytic domain but not the N-terminal RNA/DNA interaction motifs, indicating a direct effect of DNA onto the catalytic domain (Figure S1D). Furthermore, we addressed whether PARP10 is capable of catalyzing pADPr formation on heterologous substrates. As shown for a bacterially expressed catalytic fragment of PARP10 (Yu et al., 2005), all four core histones were substrates of C-TAP-PARP10 and GST-PARP10(818 1025) but did not show any obvious shift in apparent molecular weight (Figure S1E). To further compare the PARP1- and PARP10-catalyzed reactions, automodication products were analyzed directly, rst by assaying for AMP and phosphoribosyl-AMP (PRAMP) and second by determining polymer length. PRAMP represents the phosphodiesterase reaction product of pADPr, while AMP is derived from mono-ADP-ribosylated proteins and the terminal ADP-ribose of polymers. Automodied PARP1 yielded a strong PRAMP signal, but PRAMP was not detectable in the PARP10catalyzed reaction (Figure 1F), consistent with a lack of pADPr formation on automodied PARP10. In contrast, AMP was released from automodied PARP10, conrming that it is mono-ADP-ribosylated. AMP was a minor product of automodied PARP1, consistent with long polymers. Indeed, PARP1 catalyzed the formation of longer polymers with increasing NAD+ concentrations, while mainly monomeric ADP-ribose and minute amounts of di- and trimers were associated with PARP10 automodication (Figure 1G). This difference in mono- versus poly-ADP-ribose formation was reected in the considerably

(B) Indicated proteins were incubated in the presence of a constant amount of 32P-NAD+ and increasing concentrations of unlabeled NAD+ as indicated. Proteins were separated by SDS-PAGE, visualized by Coomassie staining, and incorporated label detected by autoradiography. (C) The reactions were performed as in (B) but without 32P-NAD+. The reaction products were analyzed by western blotting using polyclonal antibodies (96-10-04) or a mAb (10H) specic for pADPr. PARP1 and PARP10 were detected using specic polyclonal antibodies. (D) HEK293 cells were transfected with plasmids expressing the indicated HA-tagged proteins. These were immunoprecipitated from whole-cell lysates and their auto-ADP-ribosylation measured using 32P-NAD+ and the indicated concentrations of unlabeled NAD+. (E) Endogenous PARP10 was immunoprecipitated from U937 promyelocytes using the polyclonal serum 891-6 or the corresponding prebleed 891-0 (Yu et al., 2005) and analyzed for auto-ADP-ribosylation as in (B). (F) PARP1 and C-TAP-PARP10 were automodied under standard reaction conditions using 5 mCi of 32P-NAD+. ADP-ribosylation reaction products were subjected to phosphodiesterase treatment. Digestion products were analyzed by 2D-TLC and detected by autoradiography. (G) PARP1 and PARP10 were automodied as described in (B). Reaction products were removed from the proteins by base treatment and analyzed on a sequencing gel. Arrowheads indicate the length of ADP-ribose chains as determined by comigration of bromophenol blue (8-mer) and xylene cyanol (20-mer) (AlvarezGonzalez and Jacobson, 1987; Tanaka et al., 1978). Autoradiograms are displayed.

Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc. 59

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

Figure 2. Comparison between PARP1 and PARP10

(A) A structure-based sequence alignment of the catalytic domains of PARP1, PARP3, PARP10, and PARP12 is shown. Yellow shadings indicate structurally conserved regions. Asterisks indicate the residues of the conserved HYE triad. (B) Superposition of PARP1 (magenta) and the PARP10 model (blue); only the backbone CA atoms are shown. The expected position of bound NAD+ is indicated based on the structure of diphtheria toxin in complex with NAD+ (Bell and Eisenberg, 1996). Secondary structure elements are labeled and correspond to the structurally conserved regions indicated in (A). In addition, the side chain of I987 in PARP10 is shown; this residue is at the beginning of b-5, in the position occupied by the catalytic glutamate in PARP1. (C) Detailed view of the nicotinamide-ribose-binding site in PARP1 (magenta) and PARP10 (blue). The side chains of the HYE motif in PARP1 (residues H862, Y896, and E988) are shown, along with the corresponding residues in PARP10.

lower Vmax of PARP10 as compared to PARP1 (2 versus 500 pmol/min/mg, respectively) (Alvarez-Gonzalez, 1988). However, the KM values for NAD+, roughly 50 mM for PARP10 (Figure S1F), were comparable to the values determined for PARP1 (Mendoza-Alvarez and Alvarez-Gonzalez, 1993). In summary, the lack of detectable mobility shifts of automodied PARP10 and of ADP-ribosylated core histones, the poor reactivity with two antibodies commonly used to detect pADPr, the lack of PRAMP upon phosphodiesterase treatment, and the lack of detectable polymer formation suggest strongly that PARP10 possesses mART but not polymerase activity. Furthermore, our ndings suggest that it is not sufcient to rely solely on the activity of the above-described antibodies to evaluate polymerase versus transferase activity. Molecular Basis for the Transferase Activity of PARP10 A model for the PARP10 catalytic domain was generated using the recently determined crystal structure of PARP12 (Protein

Data Bank [PDB] ID code 2PA9; Structural Genomics Consortium, http://www.sgc. utoronto.ca/), with which it shares 33% sequence identity. A structure-based alignment of PARP1, PARP3, PARP12, and the PARP10 model is shown in Figure 2A. PARP1 has pADPr polymerase activity, while PARP10 catalyzes only mono-ADP-ribosylation. To identify the origins of this difference, we compared the PARP10 model with that of PARP1. Core secondary structure elements are retained (Figure 2B), as are the histidine (H887) and tyrosine (Y919) involved in NAD+ binding and which constitute the rst two residues of the highly conserved HYE triad found throughout the ADPr transferase superfamily (Figure 2C). However, it is clear that in PARP10 an isoleucine (I987) has replaced the catalytic glutamate found in PARP1 (E988) (Figure 2C). Therefore, even though a catalytic glutamate is present in all other related ADP-ribosyltransferases (Domenighini et al., 1994; Domenighini and Rappuoli, 1996; Okazaki and Moss, 1994), the structure of PARP12 and the PARP10 model conrm the surprising observation that the catalytic glutamate is not present in these enzymes. Perhaps more surprising is that a close inspection of the PARP10 and PARP12 structures indicates there is no other residue in the active site that could substitute for the catalytic glutamate. In addition to the absence of the catalytic glutamate in PARP10, there is a striking difference in the sequence that connects b-4 and b-5. In PARP1, this connecting loop is 37 residues long, whereas it is only 6 residues long in PARP10 (Figure 2A). In PARP1, the catalytic glutamate (E988) is positioned at the N terminus of b-5, and the length and structure of the b-4/b-5 loop makes the region around E988 relatively crowded and

60 Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc.

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

Figure 3. Three Different Classes of PARP Enzymes

The catalytic domains of the different PARP enzymes were compared according to the structural model developed here. The catalytic core motifs and the loop length (between b sheets 4 and 5) are compared. The three amino acids relevant for catalytic activities of PARP enzymes are given. Histidine and tyrosine are within the catalytic center and are important for NAD+ binding. The third amino acid, glutamate, has been demonstrated to be critical for polymerase activity of PARP1 (Marsischky et al., 1995; Rolli et al., 1997). According to the catalytic core motif and the loop length, as given in the gure; the described activities of the PARP enzymes that have been analyzed; and the ndings described here, three subfamilies are postulated. The rst consists of pADPr polymerases (yellow), the second of mono-ADP-ribosyltransferases (green), and the third of catalytically inactive PARP family members (grey). Long, intermediate, and short loops are indicated by +++, ++, and +, respectively.

constricted. In contrast, the short loop in PARP10 adopts a conformation that effectively removes it from the active site region, making it much more open and accessible when compared with PARP1 (Figure 2B). The lack of a glutamate in the catalytic site of PARP10 and the proposed structure of the b-4/b-5 loop, which has been previously implicated in substrate recognition (Han and Tainer, 2002; Ruf et al., 1998; Sun et al., 2004), appear to be the two features of PARP10 that correlate with its lack of pADPr polymerase activity. A comprehensive sequence alignment (Otto et al., 2005) indicates that PARP6PARP16 also lack a catalytic glutamate, and, in common with PARP10, most of these enzymes have a six-residue b-4/b-5 loop with a proline two residues N-terminal to the position where the catalytic glutamate is normally found. We have dened three PARP subfamilies based on the HYE triad at the catalytic core (Figure 2A, asterisks) and the length of the b-4/b-5 loop (Figure 3). The rst is characterized by an HYE triad and a long b-4/b-5 loop. This subfamily consists of PARP1 PARP5, with PARP1, PARP2, PARP5a, and PARP5b having been demonstrated to synthesize polymers. In the second group (PARP6PARP8, PARP10PARP12, and PARP14PARP16), the HYE motif is replaced by HYI/L/Y and the loop is considerably shorter. The rst member of this group that has been analyzed in detail is PARP10, reported here. Accordingly, we suggest that the members of this group function only as mARTs rather than polymerases. PARP9 and PARP13 of the third group possess QYT or YYV triads, respectively. These two PARP family members lack the catalytically important glutamate and the histidine, which binds the b-NAD+ cofactor (Ruf et al., 1996;

Ruf et al., 1998). Since PARP9 has been reported to be catalytically inactive (Aguiar et al., 2005), we suggest that PARP13 will most likely also be inactive. To evaluate these hypotheses, we tested the activities of different catalytic domains. The catalytic domain of PARP1, PARP1cat, although poorly active in comparison to the fulllength protein, was able to catalyze the formation of long pADPr polymers, as indicated by a massive shift of the ADP-ribosylated protein to higher molecular weight and strong staining with mAb 10H (Figure S2A and data not shown). The catalytic domain of PARP13, with YYV in the catalytic center and thus lacking both the glutamate and the histidine, did not reveal any auto-ADPribosylating activity (Figure S2A). PARP14 was capable of ADP-ribosylation but did not reveal any mobility shift and did not react with mAb 10H, and upon phosphodiesterase treatment only AMP but no PRAMP was released (Figures S2A and S2B and data not shown). Thus, PARP14 functions as a monoADPr transferase and not a polymerase. These ndings support the structure- and sequence-based division of the PARP family into three functionally distinct subgroups, one with pADPr polymerase activity, another with mART activity, and a third that is catalytically inactive (for summary, see Figure 3). Characterization of the Catalytic Activity of PARP10: ADP-Ribosylation of Acidic Residues To understand the transferase function of PARP10 in more detail, we analyzed its catalytic properties. The automodication of PARP10 proceeded with second-order kinetics, indicative of an intermolecular reaction (Figure 4A). In addition, the inactive

Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc. 61

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

Figure 4. PARP10 ADP-Ribosylates In trans and Targets Acidic Residues

(A) C-TAP-PARP10 was titrated as indicated and the reaction products analyzed by SDS-PAGE (left panel). The labeled bands were cut out from the dried gel and counted. The mean values and standard deviation of three experiments are summarized in the right panel. (B) The indicated PARP10 fragments were expressed as GST-fusion proteins and subjected to an enzymatic PARP assay with C-TAP-PARP10 or C-TAPPARP10-G888W. Arrowheads indicate GST and the different GST-PARP10 fusion proteins. The labeled proteins, i.e., PARP10 full length and GSTPARP10(8181025), are indicated by arrows. (C) Immobilized GST-PARP10(8181025) was automodied and then subjected to treatment with 1 M neutral NH2OH. Residual radioactivity was determined by autoradiography. (D) C-TAP-PARP10 and PARP1-E988K were immobilized on Protein A-Sepharose using specic antibodies, automodied, and subsequently subjected to PARG treatment. Residual radioactivity was determined by autoradiography. (E) Immobilized GST-PARP10(8181025) was automodied and afterwards subjected to PARG/ARH assay using the indicated GST proteins. (F) Proliferation assays in HeLa cells. The plates were stained with methylene blue. The graph summarizes the results of three independent experiments showing the mean values and the according standard deviations.

C-TAP-PARP10-G888W was a substrate of GST-PARP10(818 1025), which contains the catalytic domain, supporting the notion that the reactions occur intermolecularly (Figure 4B). In order to dene the region in PARP10 that is ADP-ribosylated, we prepared overlapping GST fusion proteins covering PARP10. When incubated in the presence of C-TAP-PARP10, the only modied fragment was GST-PARP10(8181025) (Figure 4B). Thus, this

fragment is not only catalytically active but it also contains the automodication site(s). ADP-ribosylation occurs on a number of distinct amino acids, including acidic residues, arginines, and cysteines, with distinct sensitivities of the respective chemical bond (Hassa et al., 2006). Automodied PARP10 was highly sensitive to hydroxylamine treatment, indicating an ester bond between ADP-ribose and

62 Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc.

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

acidic residues (Figure 4C). However, automodied PARP10 was insensitive to HgCl2 (Figure S3A), which cleaves Cys-ADP-ribose linkages (McDonald and Moss, 1994). meta-iodobenzylguanidine (MIBG), an inhibitor of arginine-specic mARTs (Smets et al., 1990), did not inhibit the enzymatic activity of PARP10 (Figure S3B). Thus cysteines or arginines were not modied by PARP10. mARTs can be inhibited specically by vitamin K1 (Banasik et al., 1992). In agreement with the conclusion drawn above that PARP10 is a nonclassical mART, vitamin K1 did not interfere with PARP10 function (Figure S3C). In addition, the differential sensitivity to known PARP1 inhibitors distinguishes PARP10 from PARP1, despite the strong structural similarity (Figure 2 and Figure S3C). One characteristic of the poly-ADP-ribosylation of PARP1 is its sensitivity to poly-ADP-ribose glycohydrolase (PARG) and ADPribose hydrolase 3 (ARH3), while ARH1 and ARH2 are inactive toward pADPr (Oka et al., 2006). ARH1 has been described to remove ADP-ribose from proteins modied on arginines (Takada et al., 1993). So far, no enzymatic activity has been reported for ARH2. Therefore we tested whether PARG was capable of de-ADP-ribosylating C-TAP-PARP10. Indeed, we observed a more than 2-fold reduction of automodied PARP10 after PARG treatment (Figure 4D). This suggested that PARG can remove the ADP-ribose from an acidic residue, in addition to hydrolyzing the glycosidic bond between ADP-ribose units. To test this further, we used PARP1-E988K, a PARP1 mutant with low catalytic activity restricted to mono-ADP-ribosylation (Rolli et al., 1997). Importantly, this modication was reversed by PARG, supporting the notion that PARG has a mono-ADPribosyl-protein lyase activity; this activity has previously been ascribed to an enzyme distinct from PARG (Oka et al., 1984). To expand on this observation, automodied GST-PARP10(818 1025) was treated with PARG and ARH1-3. Both PARG and ARH3, but not ARH1 and ARH2, were able to de-ADP-ribosylate GST-PARP10(8181025) (Figure 4E). Thus, in addition to their described pARPr glycohydrolase activity, PARG and ARH3 are likely able to remove the last ADP-ribose moiety from proteins ADP-ribosylated on acidic residues. Together, these ndings strengthen our conclusion that PARP10 mono-ADP-ribosylates acidic residues. To address the biological signicance of the catalytic activity of PARP10, we performed proliferation assays in HeLa cells (Figure 4F). Exogenous expression of PARP10 interfered with cell proliferation. In contrast, catalytically inactive mutants had no effect (PARP10-G888W) or slightly enhanced (PARP10-H887E) proliferation. The cyclin-dependent kinase inhibitor p27KIP1 served as positive control since it is a strong repressor of cellcycle progression. The knockdown of PARP10 using two different siRNA constructs did not affect colony formation in HeLa cells (data not shown). Together these ndings demonstrate the importance of PARP10-mediated ADP-ribosylation in cells. Identication of Glu882 as Auto-ADP-Ribosylation Site It appears that PARPs modify acidic residues, but there is limited information on the identity of PARP-modied ADP-ribosylation sites; for example, the site(s) of automodication has not been mapped in PARP1, despite considerable effort (Hassa et al., 2006). To identify potential automodication sites in PARP10,

we evaluated the positions of acidic residues in PARP10 using our structural model. Mutation of four potentially accessible acidic residues at the extreme C terminus, either alone or in combination, reduced the automodication only slightly (Figure S4A), suggesting that the C terminus is not efciently ADP-ribosylated. We then turned to four surface exposed glutamates, E866, E870, E877, and E882, and mutated these residues individually to alanine. Of these mutants, GST-PARP10(8181025)-E882A was signicantly less auto-ADP-ribosylated, while the other mutants showed no or less signicant reduction (Figure 5A). Similarly, the GST-PARP10(8181025)-E882Q was poorly automodied (data not shown). No reduction in enzymatic activity per se was observed for GST-PARP10(8181025)-E882A as determined by its ability to trans-ADP-ribosylate C-TAP-PARP10-G888W (Figure 5B). Furthermore, the trans-ADP-ribosylation of the catalytically inactive GST-PARP10(8181025)-G888W-E882A by the full-length C-TAP-PARP10 was also reduced (Figure S4B). Thus these ndings suggest that E882 is a major, but apparently not the only, acceptor site for PARP10 auto-ADP-ribosylation. To investigate the feasibility of E882 of PARP10 being the automodication site, a model of the PARP10 automodication complex was constructed (Figure 5C). In this model, a substrate PARP10 (S-PARP10) was complexed with an enzymatic PARP10-NAD+ complex (NAD-PARP10). The NAD+ was positioned based on a complex between diphtheria toxin and NAD+ (PDB ID 1TOX; Bell and Eisenberg [1996]). With relatively minor adjustments to loop regions, we were able to position the carboxylate of E882 in S-PARP10 very close to I987 of NAD-PARP10 such that the carboxylate of E882 occupies a position roughly equivalent to the position occupied by the carboxylate of the catalytic glutamate in PARP1 (Figures 5C5E). Thus the pocket between the ribose-nicotinamide of NAD+ and I987 is sufciently large to accommodate the glutamate of the substrate. Substrate-Assisted Catalysis by PARP10 The ADP-ribosylation reaction has been well-studied, and a mechanism incorporating the most important features is illustrated in Figure 6A. NAD+ is bound by the enzyme, with the nicotinamide moiety positioned in a deep pocket and isolated from solvent; the lack of support for the positive charge on the nicotinamide promotes formation of the transition state (Berti et al., 1997). For the toxins, measurement of kinetic isotope effects indicate that the transition state bond orders are very small for both the leaving nicotinamide group and the incoming nucleophile and that the transition state has a high degree of oxacarbenium ion character (Berti et al., 1997; Parikh and Schramm, 2004; Rising and Schramm, 1997; Scheuring et al., 1998; Scheuring and Schramm, 1997). In other words, when the nucleophile has begun to form a bond with C10 , the bond to the leaving group has almost been broken. Thus, for diphtheria toxin, the mechanism has been termed a dissociative Sn2 reaction (Berti et al., 1997). The catalytic glutamate does not appear to be important for binding of NAD+, since its mutation in PARP1 (Marsischky et al., 1995), as well as the bacterial toxins (Bohmer et al., 1996; Douglas and Collier, 1990), does not affect the dissociation constant for NAD+. Instead, the catalytic glutamate functions to

Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc. 63

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

Figure 5. E882 Is ADP-Ribosylated in an Intermolecular Automodication Reaction

(A) The indicated GST-fusion proteins were subjected to a standard PARP assay. Incorporated radioactivity was detected and quantied on a phosphoimager. Relative ADP-ribosylation was calculated using mean values derived from four separate experiments. Depicted is a representative experiment. (B) The indicated PARP10 fragments were expressed as GST-fusion proteins, automodied in the presence of C-TAP-PARP10-G888W, and analyzed by autoradiography. (C) Two views of the automodication complex consisting of a PARP10 molecule with bound NAD+ (surface representation, along with NAD+ as sticks with yellow carbon atoms) and a PARP10 substrate molecule (blue), including the side chain of E882, a site of ADP-ribosylation. (D) Detail of the residues surrounding the E882 site of modication; residues from the substrate are illustrated with carbon atoms colored cyan with the sequence RPVEQV. (E) The carboxylate of E882 (cyan carbons) occupies a position very close to that of the catalytic glutamate (E988) in PARP1 (magenta carbons) and can therefore act to stabilize the oxacarbenium ion transition state.

stabilize the oxacarbenium ion transition state that develops during the reaction. On this basis, the carboxyl group of the catalytic glutamate must be very close to the C10 -N bond and therefore well-positioned to stabilize the oxacarbenium ion transition state through a charge interaction. Given this well-characterized and essential function of the catalytic glutamate, its replacement with isoleucine in PARP10 raises the question of how this enzyme can efciently catalyze

ADP-ribosylation. Both PARP1 and PARP10 catalyze ADPribosylation on substrate glutamate residues. PARP1 also has a polymerase activity and catalyzes elongation and branching reactions by adding ADP-ribosyl moieties to the initial glutamate-linked ADP-ribose and then to the growing pADPr chain. The nucleophile for the initial reaction is the carboxylate of the substrate glutamate, while the nucleophiles for the polymerization reactions are ribose hydroxyls. For PARP1, mutagenesis

64 Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc.

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

Figure 6. PARP10 Uses Substrate-Assisted Catalysis

(A) The dissociative Sn2-type mechanism. In this mechanism, the transition state is stabilized by the catalytic glutamate and has a high degree of oxacarbenium ion character. (B) The indicated GST-PARP10 fragments were incubated in the presence of C-TAP-PARP10-G888W with 32P-NAD+ and 50 mM NAD+. The reaction products were analyzed by autoradiography. (C) A fully dissociative substrate-assisted catalytic mechanism. The glutamate is attached to the substrate and is inserted into a position where it can promote the oxacarbenium ion transition state and react with it once it has formed.

of the catalytic glutamate shows that these two reactionschain initiation and polymerizationhave different catalytic requirements. For example, mutation of the catalytic glutamate to alanine (E988A) abrogates polymerization, but the enzyme can still

catalyze chain initiation, although at low rates compared to wild-type. Mutation of E988 to glutamine still results in loss of polymerization activity, but the mutant enzyme has only a slightly decreased ability to catalyze chain initiation. Finally, an E988D

Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc. 65

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

Figure 7. Functional Relevance of the b4/b5 Loop for PARP10 Catalysis

(A) The PARP10 loop (gray) in comparison to the PARP1 loop (magenta) is shown relative to substrate PARP10 (cyan) with the indicated acceptor site E882. (B) The indicated GST-PARP10 fragments with mutations in the loop region were subjected to a PARP assay with C-TAP-PARP10-G888W.

mutant catalyzes both chain initiation and elongation, but at reduced rates compared to wild-type (Marsischky et al., 1995; Ruf et al., 1998). Thus, for PARP1, the degree of oxacarbenium ion stabilization by the residue at position 988 determines the catalytic properties of the enzyme. On this basis, the presence of isoleucine at the position of the catalytic glutamate in PARP10 could explain its lack of polymerase activity, and we reasoned that replacement of I987 with glutamate might restore a polymerase activity and improve its ability to catalyze chain initiation. To address whether replacing I987 by glutamate in PARP10 might be sufcient to obtain a polymerase, GST-PARP10(818 1025)-I987E was tested for catalytic activity. Unexpectedly, the I987E mutation strongly reduced mART activity without evidence for polymerase activity (Figure 6B). In contrast, mutation of T999, which was suggested to correspond to the catalytic E988 of PARP1 (Ame et al., 2004), to glutamate had no effect on catalytic activity (Figure 6B). This is consistent with its location on the surface of our PARP10 model, away from the active site (data not shown). The loss of activity in the I987E mutant is difcult to explain given the conserved and virtually essential role for a catalytic glutamate in related ADP-ribosyltransferases and also in light of the structural conservation of the enzyme, particularly in the active site region. Mutations at this position have been accommodated in many of the other related ADPr transferases, and from the PARP10 structure there is no obvious reason why a glutamic acid could not be accommodated at position 987. On this basis, we concluded that PARP10 must use a different catalytic mechanism than the other ADP-ribosyltransferases, namely one that involves substrate-assisted catalysis. For substrate-assisted catalysis to be operational in PARP10, the site of modication also fullls the function of the catalytic glutamate and must therefore occupy a position close to where the catalytic glutamate would normally reside. This is supported

by our structural model described above (Figures 5C5E). A substrate-assisted catalytic mechanism could work as follows. First, given the similarity in the structure of PARP10 with other ADPribosyltransferases, along with the identity of the NAD+ substrate, it is likely that the reaction proceeds through a similar oxacarbenium ion transition state. However, since PARP10 lacks a catalytic glutamate at position 987, the transition state is stabilized by a substrate glutamate and the reaction would proceed by a fully dissociative Sn1-type of mechanism (Figure 6C). Insertion of the substrate glutamate into the region next to the nicotinamide ribose would promote formation of the oxacarbenium ion and cleavage of the ribose-nicotinamide bond, followed by nucleophilic attack on the oxacarbenium ion. A mechanism incorporating substrate-assisted catalysis explains the surprising strong reduction of activity of the I987E mutant of PARP10: insertion of an acidic group in exactly the same region that the substrate glutamate must bind would interfere with modication of the substrate. Functional Relevance of the Conserved Loop Region between b Sheets 4 and 5 In addition to the amino acids in the catalytic center, the loop region differs between the three proposed PARP subfamilies (Figures 2A and 3). The protein model suggests that for the substrate to have access to this region, the structure of the loop connecting b strands 4 and 5 is critical (Figure 7A). In this regard, PARP6PARP8, PARP10PARP12, and PARP14PARP16 share a highly conserved feature: the loop for these proteins is very short. With the exception of PARP6 and PARP8, there is a conserved proline two residues N-terminal to the position normally occupied by the catalytic glutamate. Because of the limited rotation around the Phi bond, this proline residue places structural constraints on the polypeptide chain, and therefore the structure of the loop in the PARP12 structure is likely conserved in the

66 Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc.

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

other PARPs of this subfamily, including PARP10. Note that PARP6 and PARP8, which have an isoleucine in place of the catalytic glutamic acid, possess a short two-residue turn between b4 and b5. Thus, the PARP enzymes of this subfamily most likely have a b4-b5 loop structure that allows insertion of a substratelinked glutamate residue in place of the catalytic glutamate, which is missing from these enzymes. This structural arrangement is supported by our PARP10 dimeric model (Figure 5C). To address the relevance of the loop in PARP10, we generated several mutants in this region, including changing P985 to alanine. All these mutants had little or no catalytic activity (Figure 7B). This was obvious for the activity toward the mutated protein itself as well as the catalytically inactive PARP10-G888W, which has an unaltered E882 and surrounding amino acids. These ndings strongly indicate that indeed the loop is critical for positioning the catalytic/acceptor glutamate of the substrate in the catalytic center. Thus employing substrate-assisted catalysis together with the loop structure would make PARP10 and related enzymes more specic for their substrates, since the geometrical constraints required for modication are greater than they would be for the more conventional type of reaction seen for PARP1. Indeed, PARP1 has a large number of substrates (Hassa et al., 2006). In contrast, PARP10 is highly selective, and several PARP1 substrates tested are not modied by PARP10 (Yu et al., 2005). Conclusions Here we report that one of the novel PARP family members, PARP10, lacks polymerase activity while showing robust ADP-ribosyltransferase activity. PARP10 also lacks a catalytic glutamate that has a critical function in all of the previously characterized ADP-ribosyltransferases. Strikingly, a PARP10-I987E mutant, in which the catalytic glutamate is restored, is nearly devoid of activity. On this basis, we propose that PARP10 employs a mechanism of substrate-assisted catalysis (DallAcqua and Carter, 2000; Kosloff and Selinger, 2001), where the glutamate that is modied fullls the function of the catalytic glutamate. Such a mechanism explains why PARP10 does not function as a polymerase. Since PARP14, an additional novel PARP enzyme that lacks the catalytic glutamate, possesses enzymatic activity comparable to PARP10, i.e., ADP-ribosyltransferase activity but no detectable polymer formation, our ndings suggest that PARP10 is the prototypical enzyme of a subclass of the PARP family. Thus, while PARP1PARP5 function as polymerases, the novel PARP6PARP16 are mARTs. These lack the conserved catalytic glutamate, and we suggest that they all may use substrate-assisted catalysis for ADP-ribosylating substrate. There are two exceptions within this PARP subfamily, PARP9 and PARP13, which lack the conserved histidine in the catalytic center in addition to the glutamate. The histidine is required for NAD+ binding. Indeed, these two PARPs at least do not possess auto-ADP-ribosylation capacity. In summary, we propose to subdivide the PARP family into three subfamilies, the rst with bona de pADPr polymerase activity (PARP1PARP5), the second with mART activity (PARP6PARP8, PARP10PARP12, PARP14PARP16), and the third without catalytic activity (PARP9 and PARP13). Intracellular mono-ADP-ribosylation has been postulated as a mechanism to regulate many different aspect of cell physiol-

ogy. However, the only enzymes described to mono-ADP-ribosylate substrates in cells are different bacterial toxins, while the mammalian mARTs (ectoenzymes) appear to be localized strictly extracellularly (Aktories and Barbieri, 2005; Corda and Di Girolamo, 2003; Hassa et al., 2006; Holbourn et al., 2006). Intracellular mono-ADP-ribosylation may result from the action of PARG on substrates carrying pADPr or by inefcient catalysis of polymerases. However, there is no evidence to support either model. Finally, sirtuins have been suggested to function as mARTs, in particular SIRT4 and SIRT6 (Haigis et al., 2006). With our discovery of a mono-ADP-ribosylating class of PARP enzymes, the importance of mono-ADP-ribosylation can now be addressed in more detail. The mechanism of substrate-assisted catalysis proposed here, which uses a substrate glutamate and possibly an aspartate for catalysis, together with eight additional PARP family members potentially able to mono-ADP-ribosylate, suggests that mono-ADP-ribosylation of acidic residues might be a widespread posttranslational mechanism to control protein function. In addition, mono-ADP-ribosylation might serve as acceptor sites for PARPs, possibly expanding the functional range of PARP1PARP5 considerably. It will now be important to identify the substrates that are mono-ADP-ribosylated by PARP10 and other PARP-like mARTs to elucidate the physiological role of this modication.
EXPERIMENTAL PROCEDURES TAP from stably transfected Flp-In T-REx 293 cells (Invitrogen) was performed essentially as described before (Puig et al., 2001), with modications as indicated in the Supplemental Experimental Procedures. PARP assays were carried out in 30 ml reaction volume at 30 C for 30 min in 50 mM Tris (pH 8.0), 0.2 mM DTT, and 4 mM MgCl2 using 50 mM b-NAD+ (Sigma) and 1 mCi 32P-NAD+ (Amersham Biosciences) if not indicated otherwise. Enzyme (500 ng) and 1 mg of substrates were used routinely. PARP1 used as positive control was baculo derived. This protein was already associated with DNA resulting in full enzymatic activity, and addition of DNA did not result in a further activation (our unpublished data). The analysis of ADP-ribosylation reaction products was carried out as described previously (Panzeter and Althaus, 1990). The model of the catalytic domain of PARP10 is based on the crystal structure of PARP12 (PDB ID 2PQF; about 33% sequence identity with PARP10). In addition, we aligned the structures of human PARP1 (PDB ID 1UK0) and PARP3 (PDB ID 2PA9) with PARP12. More detailed experimental procedures can be found in the Supplemental Data.

SUPPLEMENTAL DATA The Supplemental Data include Supplemental Experimental Procedures, Supplemental References, and four gures and can be found with this article online at http://www.molecule.org/cgi/content/full/32/1/57/DC1/.

ACKNOWLEDGMENTS We thank M. Kracht for reagents, K. Montzka for help with cloning PARP10 fragments, C. Blenn for many helpful discussions, A. Szameit for excellent technical assistance, and Alexis Biochemicals for providing PARP inhibitors. This work was supported by Marie Curie Host Fellowships for Transfer of Knowledge (FUNGEN, FP6-517068) to E.P. and K.L., by Swiss National Science Foundation Grants 31-109315.05 and the Kanton of Zurich to P.O.H. and M.O.H., and by the Deutsche Forschungsgemeinschaft (LU 466/13-1) to B.L.

Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc. 67

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

Received: January 12, 2008 Revised: May 23, 2008 Accepted: August 1, 2008 Published: October 9, 2008 REFERENCES Aguiar, R.C., Takeyama, K., He, C., Kreinbrink, K., and Shipp, M.A. (2005). B-aggressive lymphoma family proteins have unique domains that modulate transcription and exhibit poly(ADP-ribose) polymerase activity. J. Biol. Chem. 280, 3375633765. Aktories, K., and Barbieri, J.T. (2005). Bacterial cytotoxins: targeting eukaryotic switches. Nat. Rev. Microbiol. 3, 397410. Alvarez-Gonzalez, R. (1988). 30 -Deoxy-NAD+ as a substrate for poly(ADP-ribose)polymerase and the reaction mechanism of poly(ADP-ribose) elongation. J. Biol. Chem. 263, 1769017696. Alvarez-Gonzalez, R., and Jacobson, M.K. (1987). Characterization of polymers of adenosine diphosphate ribose generated in vitro and in vivo. Biochemistry 26, 32183224. Ame, J.C., Rolli, V., Schreiber, V., Niedergang, C., Apiou, F., Decker, P., Muller, S., Hoger, T., Menissier-de Murcia, J., and de Murcia, G. (1999). PARP-2, A novel mammalian DNA damage-dependent poly(ADP-ribose) polymerase. J. Biol. Chem. 274, 1786017868. Ame, J.C., Spenlehauer, C., and de Murcia, G. (2004). The PARP superfamily. Bioessays 26, 882893. Banasik, M., Komura, H., Shimoyama, M., and Ueda, K. (1992). Specic inhibitors of poly(ADP-ribose) synthetase and mono(ADP-ribosyl)transferase. J. Biol. Chem. 267, 15691575. Bell, C.E., and Eisenberg, D. (1996). Crystal structure of diphtheria toxin bound to nicotinamide adenine dinucleotide. Biochemistry 35, 11371149. Bellocchi, D., Costantino, G., Pellicciari, R., Re, N., Marrone, A., and Coletti, C. (2006). Poly(ADP-ribose)-polymerase-catalyzed hydrolysis of NAD+: QM/MM simulation of the enzyme reaction. ChemMedChem 1, 533539. Berti, P.J., Blanke, S.R., and Schramm, V.L. (1997). Transition state structure for the hydrolysis of NAD+ catalyzed by diphtheria toxin. J. Am. Chem. Soc. 119, 1207912088. Bohmer, J., Jung, M., Sehr, P., Fritz, G., Popoff, M., Just, I., and Aktories, K. (1996). Active site mutation of the C3-like ADP-ribosyltransferase from Clostridium limosumanalysis of glutamic acid 174. Biochemistry 35, 282289. Chou, H.Y., Chou, H.T., and Lee, S.C. (2006). CDK-dependent activation of poly(ADP-ribose) polymerase member 10 (PARP10). J. Biol. Chem. 281, 1520115207. Corda, D., and Di Girolamo, M. (2003). Functional aspects of protein monoADP-ribosylation. EMBO J. 22, 19531958. DAmours, D., Desnoyers, S., DSilva, I., and Poirier, G.G. (1999). Poly(ADPribosyl)ation reactions in the regulation of nuclear functions. Biochem. J. 342, 249268. DallAcqua, W., and Carter, P. (2000). Substrate-assisted catalysis: molecular basis and biological signicance. Protein Sci. 9, 19. Domenighini, M., and Rappuoli, R. (1996). Three conserved consensus sequences identify the NAD-binding site of ADP-ribosylating enzymes, expressed by eukaryotes, bacteria and T-even bacteriophages. Mol. Microbiol. 21, 667674. Domenighini, M., Magagnoli, C., Pizza, M., and Rappuoli, R. (1994). Common features of the NAD-binding and catalytic site of ADP-ribosylating toxins. Mol. Microbiol. 14, 4150. Douglas, C.M., and Collier, R.J. (1990). Pseudomonas aeruginosa exotoxin A: alterations of biological and biochemical properties resulting from mutation of glutamic acid 553 to aspartic acid. Biochemistry 29, 50435049. Haigis, M.C., Mostoslavsky, R., Haigis, K.M., Fahie, K., Christodoulou, D.C., Murphy, A.J., Valenzuela, D.M., Yancopoulos, G.D., Karow, M., Blander, G., et al. (2006). SIRT4 inhibits glutamate dehydrogenase and opposes the effects of calorie restriction in pancreatic beta cells. Cell 126, 941954.

Haince, J.F., Rouleau, M., Hendzel, M.J., Masson, J.Y., and Poirier, G.G. (2005). Targeting poly(ADP-ribosyl)ation: a promising approach in cancer therapy. Trends Mol. Med. 11, 456463. Han, S., and Tainer, J.A. (2002). The ARTT motif and a unied structural understanding of substrate recognition in ADP-ribosylating bacterial toxins and eukaryotic ADP-ribosyltransferases. Int. J. Med. Microbiol. 291, 523529. Hassa, P.O., Haenni, S.S., Elser, M., and Hottiger, M.O. (2006). Nuclear ADPribosylation reactions in mammalian cells: where are we today and where are we going? Microbiol. Mol. Biol. Rev. 70, 789829. Holbourn, K.P., Shone, C.C., and Acharya, K.R. (2006). A family of killer toxins. Exploring the mechanism of ADP-ribosylating toxins. FEBS J. 273, 45794593. Kawamitsu, H., Hoshino, H., Okada, H., Miwa, M., Momoi, H., and Sugimura, T. (1984). Monoclonal antibodies to poly(adenosine diphosphate ribose) recognize different structures. Biochemistry 23, 37713777. Kim, M.Y., Zhang, T., and Kraus, W.L. (2005). Poly(ADP-ribosyl)ation by PARP-1: PAR-laying NAD+ into a nuclear signal. Genes Dev. 19, 19511967. Kosloff, M., and Selinger, Z. (2001). Substrate assisted catalysisapplication to G proteins. Trends Biochem. Sci. 26, 161166. Marsischky, G.T., Wilson, B.A., and Collier, R.J. (1995). Role of glutamic acid 988 of human poly-ADP-ribose polymerase in polymer formation. Evidence for active site similarities to the ADP-ribosylating toxins. J. Biol. Chem. 270, 32473254. McDonald, L.J., and Moss, J. (1994). Enzymatic and nonenzymatic ADPribosylation of cysteine. Mol. Cell. Biochem. 138, 221226. Mendoza-Alvarez, H., and Alvarez-Gonzalez, R. (1993). Poly(ADP-ribose) polymerase is a catalytic dimer and the automodication reaction is intermolecular. J. Biol. Chem. 268, 2257522580. Naegeli, H., Loetscher, P., and Althaus, F.R. (1989). Poly ADP-ribosylation of proteins. Processivity of a post-translational modication. J. Biol. Chem. 264, 1438214385. Oka, J., Ueda, K., Hayaishi, O., Komura, H., and Nakanishi, K. (1984). ADPribosyl protein lyase. Purication, properties, and identication of the product. J. Biol. Chem. 259, 986995. Oka, S., Kato, J., and Moss, J. (2006). Identication and characterization of a mammalian 39-kDa poly(ADP-ribose) glycohydrolase. J. Biol. Chem. 281, 705713. Okazaki, I.J., and Moss, J. (1994). Common structure of the catalytic sites of mammalian and bacterial toxin ADP-ribosyltransferases. Mol. Cell. Biochem. 138, 177181. Otto, H., Reche, P.A., Bazan, F., Dittmar, K., Haag, F., and Koch-Nolte, F. (2005). In silico characterization of the family of PARP-like poly(ADP-ribosyl)transferases (pARTs). BMC Genomics 6, 139. Panzeter, P.L., and Althaus, F.R. (1990). High resolution size analysis of ADPribose polymers using modied DNA sequencing gels. Nucleic Acids Res. 18, 2194. Parikh, S.L., and Schramm, V.L. (2004). Transition state structure for ADPribosylation of eukaryotic elongation factor 2 catalyzed by diphtheria toxin. Biochemistry 43, 12041212. Puig, O., Caspary, F., Rigaut, G., Rutz, B., Bouveret, E., Bragado-Nilsson, E., Wilm, M., and Seraphin, B. (2001). The tandem afnity purication (TAP) method: a general procedure of protein complex purication. Methods 24, 218229. Rippmann, J.F., Damm, K., and Schnapp, A. (2002). Functional characterization of the poly(ADP-ribose) polymerase activity of tankyrase 1, a potential regulator of telomere length. J. Mol. Biol. 323, 217224. Rising, K.A., and Schramm, V.L. (1997). Transition state analysis of NAD+ hydrolysis by the cholera toxin catalytic subunit. J. Am. Chem. Soc. 119, 2737. Rolli, V., OFarrell, M., Menissier-de Murcia, J., and de Murcia, G. (1997). Random mutagenesis of the poly(ADP-ribose) polymerase catalytic domain reveals amino acids involved in polymer branching. Biochemistry 36, 12147 12154.

68 Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc.

Molecular Cell
PARP10 Functions as a Mono-ADP-Ribosyltransferase

Ruf, A., Mennissier de Murcia, J., de Murcia, G., and Schulz, G.E. (1996). Structure of the catalytic fragment of poly(AD-ribose) polymerase from chicken. Proc. Natl. Acad. Sci. USA 93, 74817485. Ruf, A., Rolli, V., de Murcia, G., and Schulz, G.E. (1998). The mechanism of the elongation and branching reaction of poly(ADP-ribose) polymerase as derived from crystal structures and mutagenesis. J. Mol. Biol. 278, 5765. Scheuring, J., and Schramm, V.L. (1997). Kinetic isotope effect characterization of the transition state for oxidized nicotinamide adenine dinucleotide hydrolysis by pertussis toxin. Biochemistry 36, 45264534. Scheuring, J., Berti, P.J., and Schramm, V.L. (1998). Transition-state structure for the ADP-ribosylation of recombinant Gialpha1 subunits by pertussis toxin. Biochemistry 37, 27482758. Schreiber, V., Dantzer, F., Ame, J.C., and de Murcia, G. (2006). Poly(ADPribose): novel functions for an old molecule. Nat. Rev. Mol. Cell Biol. 7, 517528. Smets, L.A., Loesberg, C., Janssen, M., and Van Rooij, H. (1990). Intracellular inhibition of mono(ADP-ribosylation) by meta-iodobenzylguanidine: specic-

ity, intracellular concentration and effects on glucocorticoid-mediated cell lysis. Biochim. Biophys. Acta 1054, 4955. Smith, S., Giriat, I., Schmitt, A., and de Lange, T. (1998). Tankyrase, a poly(ADPribose) polymerase at human telomeres. Science 282, 14841487. Sun, J., Maresso, A.W., Kim, J.J., and Barbieri, J.T. (2004). How bacterial ADPribosylating toxins recognize substrates. Nat. Struct. Mol. Biol. 11, 868876. Takada, T., Iida, K., and Moss, J. (1993). Cloning and site-directed mutagenesis of human ADP-ribosylarginine hydrolase. J. Biol. Chem. 268, 17837 17843. Tanaka, M., Hayashi, K., Sakura, H., Miwa, M., Matsushima, T., and Sugimura, T. (1978). Demonstration of high molecular weight poly (adenosine diphosphate ribose). Nucleic Acids Res. 5, 31833194. Yu, M., Schreek, S., Cerni, C., Schamberger, C., Lesniewicz, K., Poreba, E., tzinger, J., Kremmer, E., et al. (2005). Vervoorts, J., Walsemann, G., Gro PARP-10, a novel Myc-interacting protein with poly(ADP-ribose) polymerase activity, inhibits transformation. Oncogene 24, 19821993.

Molecular Cell 32, 5769, October 10, 2008 2008 Elsevier Inc. 69